In theory: What's the best approach to a multi-phase widening project?

[wanderer2575]

The scenario:  A 50-mile stretch of freeway is being widened from two lanes each direction to three.  Let's say the freeway is already three lanes each direction on either side of this project stretch so there will be no lane reduction after completion.  Due to budget or other constraints only one five-mile segment will be completed each year, so it's a 10-year project.  What's the ideal way of ordering the segments?

Start from the project ends and work toward the center?
For example:
          _______ _______ _______ _______ _______ _______ _______ _______ _______ _______
Year         1           2          5           8           9          10          7          6          4          3

Or do every other segment and then go back to fill the gaps?
For example:
          _______ _______ _______ _______ _______ _______ _______ _______ _______ _______
Year         1          6           2            7          3          8           4          9           5          10

Would it matter if the number of segments and/or length of each segment was different?

[webny99]

You forgot one of the most obvious options: A nice tidy progression, starting at one end and working towards the other.

Personally, I'd just use AADT data and widen the busiest sections first.

[jmacswimmer]

Coincidentally, I-64 between Richmond & Newport News fits the OP description almost perfectly.

Just over 50 miles from the I-295 interchange to where the road formerly narrowed to 4 lanes west of exit 255.  The work has generally started from this eastern point (first 2 phases complete, with the 3rd currently under construction) and is working west, with the exception being the recent 6-laning between exits 200 & 205 at the west end.  Since AADT picks up at Williamsburg heading east, it made sense to start at the east end.

[wanderer2575]

You forgot one of the most obvious options: A nice tidy progression, starting at one end and working towards the other.

Essentially the same as starting at the ends and working in, but I see your point. 

Rewording my question:  Better to keep one contiguous unwidened section, or several discontiguous widened and unwidened segments?  Would opening a new lane and then ending it, every five miles, screw up traffic flow?

[sprjus4]

Coincidentally, I-64 between Richmond & Newport News fits the OP description almost perfectly.

Just over 50 miles from the I-295 interchange to where the road formerly narrowed to 4 lanes west of exit 255.  The work has generally started from this eastern point (first 2 phases complete, with the 3rd currently under construction) and is working west, with the exception being the recent 6-laning between exits 200 & 205 at the west end.  Since AADT picks up at Williamsburg heading east, it made sense to start at the east end.

I'll also add that it also made sense to get that western most part done between Exit 200 and Exit 205 early on, which they did, completed just months after Phase 2 was down on the Peninsula, because it expanded out that merge area that often caused massive delays from I-295. Now, congestion is practically non existent there because traffic levels out across all 3 lanes first then can smoothly merge into 2. Obviously, 3 lanes should continue for good there, but the current merge point works better than the previous - actually said "merge point"  is the right lane becoming exit only at VA-156 (Exit 205) and vice versa going west at I-295 (Exit 200).

I-64 is not 6 lanes west of this example, but the traffic counts cut in half from 60,000 to just around 30,000 because of everyone heading north on I-295.

[sprjus4]

You forgot one of the most obvious options: A nice tidy progression, starting at one end and working towards the other.

Essentially the same as starting at the ends and working in, but I see your point. 

Rewording my question:  Better to keep one contiguous unwidened section, or several discontiguous widened and unwidened segments?  Would opening a new lane and then ending it, every five miles, screw up traffic flow?

On a busy corridor, yes.

[jeffandnicole]

Start on one end, and continue towards the other end. Usually, start on the busiest end first.

Note however, you don't necessarily have to complete one stretch before you start another stretch. Construction can overlap. For example if each stretch of highway will take three years to construct, you can start the first 5 miles this year, then the next 5 miles next year, then another 5 miles a year later, etc. It will take over a decade to complete, but you're putting out to bid the project every year, dividing the money up over time, and allowing more contractors the opportunity to bid various aspects of the overall job.

[michravera]

The scenario:  A 50-mile stretch of freeway is being widened from two lanes each direction to three.  Let's say the freeway is already three lanes each direction on either side of this project stretch so there will be no lane reduction after completion.  Due to budget or other constraints only one five-mile segment will be completed each year, so it's a 10-year project.  What's the ideal way of ordering the segments?

Start from the project ends and work toward the center?
For example:
          _______ _______ _______ _______ _______ _______ _______ _______ _______ _______
Year         1           2          5           8           9          10          7          6          4          3

Or do every other segment and then go back to fill the gaps?
For example:
          _______ _______ _______ _______ _______ _______ _______ _______ _______ _______
Year         1          6           2            7          3          8           4          9           5          10

Would it matter if the number of segments and/or length of each segment was different?

The answer will depend upon what factors you are trying to maximize or minimize. If you are trying to maximize throughput for the entire system, my hunch is that you find the sections that just follow those with the highest peak hour traffic counts and widen them first. If, however, this is a "nobody is here who really wants to be" section of road, you're probably best off just picking an end and working to the other. My sense of the problem (which in math is a system of differential equations) is that the solution that is best for the entire project may not be same the one that offers the most relief the soonest. It's hard to know without working the full problem.

This would be an interesting case for a simulation. (It's a rather complicated system of differential equations to solve exactly.) It may be that what improves things the most over the 10-year expansion doesn't improve things as much right away. You can get close by running your traffic model (including expected traffic increase or decrease and latent demand) at the end of each phase, but I don't know any widening phase that actually works like that. Once widening is completed on a portion of the road, it usually gets opened. So, it's more reasonable and likely that you'd open up a half-mile section in each direction every 17 or 18 days. There will definitely be some feedback in the loop.

[GaryV]

One consideration could be bridges.  You can't knock down every bridge crossing the freeway at the same time, or there would be no way for emergency vehicles to get through.

When they widened I-75 to the SE of Pontiac, they painted out the new lane with yellow stripes.  The roadwork was done, but can't be used until other parts are completed.

[Mr_Northside]

Another factor could be if any stretch has any elements that need replacing more urgently.   Like, if there were some bridges that, for whatever reason, were distinctly in worse shape than other structures in a particular section - you might want to get that section done first.

[Scott5114]

Texas would widen all 50 miles all at once.

Oklahoma would widen one mile at a time for the next 50 years.

[paulthemapguy]

You will not see a state government allocate the funds to widen 50 miles of freeway all at the same time.  This might as well be in fictional highways.  States will widen the freeway in the order that funding is scrounged up for it.

[sprjus4]

Texas would widen all 50 miles all at once.

Oklahoma would widen one mile at a time for the next 50 years.

Sounds accurate.

[sprjus4]

You will not see a state government allocate the funds to widen 50 miles of freeway all at the same time.  This might as well be in fictional highways.  States will widen the freeway in the order that funding is scrounged up for it.

Texas has been punching out 20-30 miles at a time for I-10 heading west of Houston.

[sparker]

Out here, Caltrans seems to prefer 7-10 mile chunks at a time -- this is the approach D6 has used for the stretch of CA 99 between CA 198 and Selma, where the longstanding 6-lane section into Fresno commences -- as well as a section of road that has seen almost continuous construction efforts for the last 25 years.  However, if it involves a significant structure, such as the Kings River bridge(s), the specific let project usually will involve the structure itself plus approaches -- a much smaller total length -- in order to keep the line expense for the project within district budget constraints.   Since most 4-to-6 expansion involves median infill, often the bridges (particularly in northern Tulare County, notorious for underheight overcrossings) are not addressed in the initial widening contract; D6 seems to procrastinate on those as long as it can.  But Caltrans, in general, tends to do whatever teardown of existing shoulders, median features, and anything else in the way all at once, then start the actual widening in a counterdirection from the flow of traffic -- i.e., on northbound lanes, they'll start at the north end and work their way south, placing shoulders and/or barriers so that the lane can be used once done; when they get to the other end, they do the same in the other direction -- except now the barrier is already in place, so it progresses faster.  Of course, if they did it the other way they would be forcing traffic to merge right in the middle of a construction zone -- a recipe for trouble.